def align_reads(fastq1, fastq2):
v3loop_ref = extract_v3loop_ref()
reader = FastqReader(fastq1, fastq2)
merged_reads = merge_reads(reader)
score_counts = Counter()
for _ in trim_reads(merged_reads, v3loop_ref, score_counts):
pass
return score_counts
def align_reads(fastq1, fastq2):
v3loop_ref = extract_v3loop_ref()
reader = FastqReader(fastq1, fastq2)
merged_reads = merge_reads(reader)
score_counts = Counter()
for _ in trim_reads(merged_reads, v3loop_ref, score_counts):
pass
return score_counts
def test_no_overlap(self):
reads = [("A:B:C", ("X:Y", "AAACCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "ACACACACACAC", "BBBBBBBBBBBB"))]
expected_merged_reads = [
("A:B:C", ("X:Y", "AAACCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "ACACACACACAC", "BBBBBBBBBBBB"), None)
]
merged_reads = list(merge_reads(reads))
self.assertEqual(expected_merged_reads, merged_reads)
def test_low_quality(self):
reads = [("A:B:C", ("X:Y", "AAACCCTTTGGGAAA", "B!BBBBBBBBBBBBB"),
("Q:R", "GGGTTTCCCAAA", "BBBBBBBBBBBB"))]
expected_merged_reads = [
("A:B:C", ("X:Y", "AAACCCTTTGGGAAA", "B!BBBBBBBBBBBBB"),
("Q:R", "GGGTTTCCCAAA", "BBBBBBBBBBBB"), "ANACCCTTTGGGAAACCC")
]
merged_reads = list(merge_reads(reads))
self.assertEqual(expected_merged_reads, merged_reads)
def test_disagreement(self):
reads = [("A:B:C", ("X:Y", "AAACCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "GGGTTTCACAAA", "@@@@@@@Y@@@@"))]
expected_merged_reads = [
("A:B:C", ("X:Y", "AAACCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "GGGTTTCACAAA", "@@@@@@@Y@@@@"), "AAACCCTTTGTGAAACCC")
]
merged_reads = list(merge_reads(reads))
self.assertEqual(expected_merged_reads, merged_reads)
def test_reverse_overlap(self):
""" The start of read 1 aligns with the end of read 2. Reject! """
reads = [("A:B:C", ("Q:R", "TTTGGGAAACCC", "BBBBBBBBBBBB"),
("X:Y", "TTTCCCAAAGGGTTT", "BBBBBBBBBBBBBBB"))]
expected_merged_reads = [
("A:B:C", ("Q:R", "TTTGGGAAACCC", "BBBBBBBBBBBB"),
("X:Y", "TTTCCCAAAGGGTTT", "BBBBBBBBBBBBBBB"), None)
]
merged_reads = list(merge_reads(reads))
self.assertEqual(expected_merged_reads, merged_reads)
def test_no_overlap(self):
reads = [("A:B:C",
("X:Y", "AAACCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "ACACACACACAC", "BBBBBBBBBBBB"))]
expected_merged_reads = [("A:B:C",
("X:Y", "AAACCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "ACACACACACAC", "BBBBBBBBBBBB"),
None)]
merged_reads = list(merge_reads(reads))
self.assertEqual(expected_merged_reads, merged_reads)
def test_low_quality(self):
reads = [("A:B:C",
("X:Y", "AAACCCTTTGGGAAA", "B!BBBBBBBBBBBBB"),
("Q:R", "GGGTTTCCCAAA", "BBBBBBBBBBBB"))]
expected_merged_reads = [("A:B:C",
("X:Y", "AAACCCTTTGGGAAA", "B!BBBBBBBBBBBBB"),
("Q:R", "GGGTTTCCCAAA", "BBBBBBBBBBBB"),
"ANACCCTTTGGGAAACCC")]
merged_reads = list(merge_reads(reads))
self.assertEqual(expected_merged_reads, merged_reads)
def test_disagreement(self):
reads = [("A:B:C",
("X:Y", "AAACCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "GGGTTTCACAAA", "@@@@@@@Y@@@@"))]
expected_merged_reads = [("A:B:C",
("X:Y", "AAACCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "GGGTTTCACAAA", "@@@@@@@Y@@@@"),
"AAACCCTTTGTGAAACCC")]
merged_reads = list(merge_reads(reads))
self.assertEqual(expected_merged_reads, merged_reads)
def test_reverse_overlap(self):
""" The start of read 1 aligns with the end of read 2. Reject! """
reads = [("A:B:C",
("Q:R", "TTTGGGAAACCC", "BBBBBBBBBBBB"),
("X:Y", "TTTCCCAAAGGGTTT", "BBBBBBBBBBBBBBB"))]
expected_merged_reads = [("A:B:C",
("Q:R", "TTTGGGAAACCC", "BBBBBBBBBBBB"),
("X:Y", "TTTCCCAAAGGGTTT", "BBBBBBBBBBBBBBB"),
None)]
merged_reads = list(merge_reads(reads))
self.assertEqual(expected_merged_reads, merged_reads)
def test_multiple_reads(self):
reads = [("A:B:C", ("X:Y", "AAACCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "GGGTTTCCCAAA", "BBBBBBBBBBBB")),
("A:B:E", ("X:Y", "TTTCCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "GGGTTTCCCAAA", "BBBBBBBBBBBB"))]
expected_merged_reads = [
("A:B:C", ("X:Y", "AAACCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "GGGTTTCCCAAA", "BBBBBBBBBBBB"), "AAACCCTTTGGGAAACCC"),
("A:B:E", ("X:Y", "TTTCCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "GGGTTTCCCAAA", "BBBBBBBBBBBB"), "TTTCCCTTTGGGAAACCC")
]
merged_reads = list(merge_reads(reads))
self.assertEqual(expected_merged_reads, merged_reads)
def test_multiple_reads(self):
reads = [("A:B:C",
("X:Y", "AAACCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "GGGTTTCCCAAA", "BBBBBBBBBBBB")),
("A:B:E",
("X:Y", "TTTCCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "GGGTTTCCCAAA", "BBBBBBBBBBBB"))]
expected_merged_reads = [("A:B:C",
("X:Y", "AAACCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "GGGTTTCCCAAA", "BBBBBBBBBBBB"),
"AAACCCTTTGGGAAACCC"),
("A:B:E",
("X:Y", "TTTCCCTTTGGGAAA", "BBBBBBBBBBBBBBB"),
("Q:R", "GGGTTTCCCAAA", "BBBBBBBBBBBB"),
"TTTCCCTTTGGGAAACCC")]
merged_reads = list(merge_reads(reads))
self.assertEqual(expected_merged_reads, merged_reads)